How Apple fell behind in the AI race

By Julia Press and Sarah Holder, Bloomberg

Over the past year, Apple has pulled out all the stops to tout shiny new AI tools: from big presentations at its Worldwide Developers Conference to ads featuring The Last of Us star Bella Ramsey. Now, the company is facing questions about what it’s promised versus what it’s delivered.

On the Big Take podcast, Bloomberg’s Mark Gurman takes host Sarah Holder inside the company’s efforts to keep up on AI and what it needs to do next to stay in the game.

Trump asks Apple to stop moving iPhone production to India

Apple may owe you money. Here’s how to get it

Google unveils big Android redesign ahead of iPhone overhaul

From Apple to GM, tariffs to cost companies tens of billions

Letters: Trump’s AmeriCorps cuts are not making America great

Here is a lightly edited transcript of the conversation:

Sarah Holder: When Apple released the iPhone 16 last year, the company put out a series of TV ads promising shiny new AI tools. One of those ads featured Bella Ramsey, the star of HBO’s The Last of Us, trying to remember someone’s name at a party.

Apple ad: Siri, what’s the name of the guy I had a meeting with a couple of months ago at Cafe Grenel?

Holder: Siri scans their calendar and answers…

Apple ad: You met Zac Wingate at Cafe Grenel…

Holder: And when Zac approaches Bella, they nail his name.

Apple ad: Hey, Zac. Oh Bella, I didn’t think you’d remember me. Yeah, of course.

Holder: It was a promise of what was to come with Apple and its AI ambitions. But Mark Gurman, who edits Bloomberg’s consumer tech coverage, and has been covering Apple for years, says… it hasn’t quite gone according to plan.

Gurman: They advertised that it was gonna do that in order to sell the new phones, but that feature never came out. A complete disconnect between Apple Engineering and Apple Marketing.

Holder: How rare is that for Apple to do something like that, to promise something, to advertise it, and then not actually deliver it?

Gurman: This is AI. This is Siri. This is at the very core of this major technological revolution. So to the scale this happened, with the importance of these features, nothing like that has happened in modern Apple history, and I consider modern Apple history to be the last 20 or so years.

Holder: Apple ended up taking the Siri ad down. But that disconnect led customers to file class action lawsuits alleging false advertising in March. Apple declined to comment on the lawsuits. The company also declined to comment on Mark’s story or on behalf of the executives mentioned.

Mark spoke with several employees and people close to the company, some of whom requested anonymity to discuss sensitive matters. And he says, based on his reporting, those missing features on the iPhone 16 point to a much bigger issue for Apple: that when it comes to the AI race, the company known for delivering on revolutionary tech is way behind.

This is the Big Take from Bloomberg News. I’m Sarah Holder.

Today on the show: inside Apple’s efforts to catch up on AI. The challenges the company faces to keep its status as a tech pioneer… and the pitfalls of getting in the game too late.

Mark, I wanna start by getting a sense of your reporting process here. What made you wanna dig into Apple’s artificial intelligence efforts?

Gurman: You know, AI has always been an important topic, but until ChatGPT launched at the end of 2022, it really didn’t come into the mainstream. It really wasn’t the center of the technology world. And it’s so interesting because over the years, Apple has dominated so many categories that it wasn’t first to: The MP3 player with the iPod, the smartphone with the iPhone, the tablet with the iPad earbuds with the AirPods, smart watches with the Apple Watch. And what was different this time around is not only was Apple late to AI or generative AI, this modern technology that we know from ChatGPT and Gemini, Anthropic, you name it. But they also weren’t the best.

There was no Apple iPhone or Apple iPad moment for AI where they took something that people didn’t really understand and made it mainstream into some beautiful, fully functional product, right? That just didn’t happen. And so for me. That was fascinating. That was a sea change for Apple.

And then over time, you start hearing from people working at Apple, people in the industry, that there’s a problem there.

Holder: What products does Apple have that do use AI today? Like when you go on your phone, is AI there?

Gurman: Touch ID, Face ID, the way you unlock your phone with biometrics. That is a form of artificial intelligence. The ability for the phone to say ‘you have a meeting in 45 minutes, there’s 40 minutes of traffic, you should probably leave right about now in order to get there on time.’ That’s artificial intelligence.

They’ve been really good at heavily integrated AI. Where they missed was this new topic of generative AI.

And so there is a big disconnect between the AI that Apple has long offered and the AI that both Wall Street and consumers are clamoring for. And Apple knew that. That’s why they spun together Apple Intelligence.

They called it “AI for the rest of us,” just like they called the original Mac, the computer for the rest of us. Expectations were sky high. The presentation looked pretty good. In reality, it fell extraordinarily flat.

I used the first beta version of Apple Intelligence back at the end of July, early August of last year. And I wrote a column about this saying ‘this is kind of unbelievable. They’ve hyped it and hyped it and hyped it, and it has basically nothing. People are gonna start using this thing and be like, that’s it?’ Right? And people were shocked at the time.

Holder: This is where you get a couple texts from your friends and then they give you basically an AI summary of what was said.

Gurman: That is one of the features. So you have the summaries and it can summarize, you know, a slew of text messages. It was able to summarize news headlines, right? But they had to pull the news headlines feature because the BBC complained that they sent a headline out about Luigi Mangione and the headline actually spit out after going through the Apple system that he had shot himself. And so that was a sign, the system was quite broken. So they pulled that months ago, and that’s still not back actually. There’s the Genmojis feature where you can create your own emoji.

Holder: Love that one.

Gurman: That is a cool feature. There’s writing tools, which allows you to summarize, text, uh, synthesize text into bullet points, but the generative AI to create something that actually uses Open AI, ChatGPT, which is also integrated into iOS 18.

So there’s a slew of these little features throughout. But many of them have also been delayed. Many of them don’t work as intended. Many of them don’t work as it’s been marketed. And what we have today is really a far cry from the vision Apple presented. And it’s an even farther cry from what you’re seeing from competitors.

Holder: Well, benefit of the doubt for a second, being a little late to the game isn’t exactly new for Apple. They’ve historically, you know, sat back while their competitors developed riskier new products. They’ve entered the ring when the bumps and the kinks are, are kind of smoothed out. Is Apple lagging behind now as a strategy to, to work more on the tech? Or is it really struggling to keep up?

Gurman: Well, I think it’s all of those things, right? One, they’re struggling to keep up. They have fewer AI engineers than other companies like, like Amazon at this point. The other issue is that they don’t have the vision for exactly how they can be different and how they can implement these things.

But also, AI is something that the company is not necessarily built to produce. AI is messy. There’s a frequent problem called hallucinations, right? Hallucinations could be you ask ChatGPT or Claude or Perplexity a question, and it’s so confident that it knows the answer, the AI, and it’ll give you an answer, but it’s complete nonsense based on nothing and it’s completely wrong.

And so Apple, as a company with 2.35 billion devices out there, they wanna avoid those types of issues. So, there is a bit of approach to go slow. There are the technical challenges that they’ve had trouble overcoming, but then there’s also the true reality that this stuff takes a lot of time in the oven in order to be at a great place for consumers and they put it in the oven quite late.

Holder: Well, let’s talk about when they put it in the oven because part of that beginning of the baking process of AI, if you will, started with poaching John Giannandrea from Google back in 2018. They wanted him to kind of kickstart the AI program at Apple. How was he supposed to change the game?

Gurman: So that was a big coup for Apple. That was one of the most dramatic hires at the time. JG, as he’s known, was probably the second most important person at Google. He ran all of Google search and all of Google AI. And don’t forget, back in 2018, Google is really at the forefront of AI, putting it into Gmail, Translate, Photos. They were really a pioneer.

And JG was supposed to come in and take everything AI-related, Siri-related, put it under his own umbrella. Before, you had Siri and different AI teams scattered throughout the corporation. Apple executives at the time felt like the scattered nature of the AI work made it more difficult for them to get things working properly. They brought it in under one roof.

He removed the head of Siri at the time. He did a lot of analysis of what features people were using and not using in Siri, and proposed killing a lot of those features. He brought in his own people from Google and elsewhere, some of the top scholars and AI researchers in the world, but then everything sort of fell flat.

Since he came to Apple, there wasn’t a lot of change that we’ve seen in Siri or Apple’s machine learning or artificial intelligence work. A lot of the AI work in the years before Apple Intelligence went to development of a self-driving car, they spent billions, billions on that. They never launched the car. That AI didn’t go entirely to waste ’cause they were able to use some of that technology towards the generative models that they’re putting on the iPhone, iPad, and Mac at this point. But not a lot happened.

Holder: Until November 2022 – when OpenAI released ChatGPT to the public. According to people familiar with the events who spoke with Mark… that set off a flurry of activity at the company…

Gurman: And Craig Federighi, who runs software engineering for Apple, he and JG and other people at Apple, they started meeting with OpenAI, met with Anthropic, met with other smaller AI players and determined they need to figure out these AI models and they need to make the 2024 release of iOS, very much an AI-driven release with AI features throughout. The edict: get as many AI features into the operating system as possible.

Holder: So three years and several delayed AI products later, the question is: when will Apple catch up? Can it? That’s after the break.

Holder: Mark, your reporting shows that internally Apple is really worried that falling behind on AI could be a critical error. But why couldn’t Apple just be content to be a good hardware and software company without being a leader in AI?

Gurman: That’s a good question. So really there’s this predicament inside Apple right now. How much of this stuff should we be building versus how much of this stuff should we be licensing? And already you have Open AI, ChatGPT integration, into Siri and writing tools for those generative use cases like writing an essay and whatnot.

They’re going to add Google Gemini as an alternative to ChatGPT inside of Siri and writing tools as well. They’re also working to redo the search engine in their browser Safari to integrate AI engines. That’s still to come. So you have this question: internal versus external partnerships. Like you said, why do we need to be an AI expert? Why can’t we just license? That’s what Samsung does, right? Samsung uses Google Gemini to power all their AI.

Holder: Right. And all these other companies are putting so many resources and energy—

Gurman: Correct.

Holder: Into developing this AI. They’re ahead.

Gurman: Correct. Sitting here today, AI is the most core fundamental technology that you can get. It’s equivalent to the processors that go into their devices. Throughout Apple’s history, it has been core technologies that have enabled their new types of products.

The iPhone only was created because they owned a core technology known as multitouch. We take it for granted today, but that touchscreen interface, the original iPhone and iPad, is enabled by very intense, expensive to develop multitouch technology . All the Macs, the one you’re using now, the iPad, those products, AirPods, were enabled by these very advanced processors.

But Apple needs to think about the next wave of technology. They already killed the self-driving car, but let’s just put that in there. So the next wave of hardware in the technology industry: autonomous cars, advanced augmented reality glasses, glasses that can scan your surrounding environment, robots, whether that’s humanoids, whether that’s roaming robots, whether that’s tabletop robots.

The only way to enable those products is by owning the core technology of AI. And we’ve already seen Apple’s AI was not up to snuff enough to produce the autonomous car, but they’re gonna be doomed on the next phase of hardware if they don’t get the AI working. And you cannot rely on third parties for technology as core as artificial intelligence. So that’s why they need to keep digging in and building their own AI to enable the next wave of hardware. ‘Cause don’t forget, at the end of the day, they’re a hardware company.

Holder: Are these things that customers are actually like really, really asking for?

Gurman: I mean, it’s hard to say. I think there is demand for augmented reality glasses. I think the Meta Ray-Bans have been somewhat popular, and so I think glasses are going to become a real category. I think there is going to be a time when pointing your watch at something or pointing your earbuds at something to get more data based on AI is going to be commonplace. I think there is going to be a market for different robotics devices and certainly, the ship has sailed. Autonomy and self-driving cars is a real thing. So I think, yes. Now, is it ever gonna be as popular as the iPhone has been over the last 20 years? Probably not, but it is certainly the future.

Holder: What does Apple actually need to change about its culture, its processes, its core business model in order to actually compete in the AI space, and is it doing it?

Gurman: Apple needs to get a lot faster. They need to get a little messier. They need to make bolder bets. They need to be less afraid to launch things. They need to go back to that ethos of move fast and break things.

There’s going to be a new entrant that potentially could knock Apple off, you know, the top of the technology mountain, right? And in order for Apple to avoid that, they’re gonna have to beat out those new entrants time and time again. And AI is the big thing right now, and they have so far very much failed to do so. Because of their large user base, because of their design, because of their marketing and the love that people have for Apple products, I mean, we’re all using them, right? They have a very big chance of turning things around, but they’re only going to have so many chances and only so much time to break through these new, faster, cheaper competitors.

Holder: Well thank you so much, Mark.

Gurman: Thanks for having me.

More stories like this are available on bloomberg.com